[3.13] gh-126742: Add _PyErr_SetLocaleString, use it for gdbm & dlerror messages (GH-126746) (GH-128023)

- Add a helper to set an error from locale-encoded `char*`
- Use the helper for gdbm & dlerror messages
(cherry picked from commit 7303f06846)

Co-authored-by: Bénédikt Tran <10796600+picnixz@users.noreply.github.com>
Co-authored-by: Victor Stinner <vstinner@python.org>
This commit is contained in:
Miss Islington (bot) 2024-12-17 12:45:29 +01:00 committed by GitHub
parent 367ea89384
commit fe08cdf265
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 176 additions and 69 deletions

View file

@ -125,6 +125,18 @@ PyAPI_FUNC(void) _PyErr_SetString(
PyObject *exception, PyObject *exception,
const char *string); const char *string);
/*
* Set an exception with the error message decoded from the current locale
* encoding (LC_CTYPE).
*
* Exceptions occurring in decoding take priority over the desired exception.
*
* Exported for '_ctypes' shared extensions.
*/
PyAPI_FUNC(void) _PyErr_SetLocaleString(
PyObject *exception,
const char *string);
PyAPI_FUNC(PyObject*) _PyErr_Format( PyAPI_FUNC(PyObject*) _PyErr_Format(
PyThreadState *tstate, PyThreadState *tstate,
PyObject *exception, PyObject *exception,

View file

@ -1,7 +1,12 @@
import _ctypes
import os import os
import sys
import unittest
import platform import platform
import sys
import test.support
import unittest
from ctypes import CDLL, c_int
from ctypes.util import find_library
FOO_C = r""" FOO_C = r"""
#include <unistd.h> #include <unistd.h>
@ -26,7 +31,7 @@ void *foo(void)
@unittest.skipUnless(sys.platform.startswith('linux'), @unittest.skipUnless(sys.platform.startswith('linux'),
'Test only valid for Linux') 'test requires GNU IFUNC support')
class TestNullDlsym(unittest.TestCase): class TestNullDlsym(unittest.TestCase):
"""GH-126554: Ensure that we catch NULL dlsym return values """GH-126554: Ensure that we catch NULL dlsym return values
@ -53,14 +58,6 @@ class TestNullDlsym(unittest.TestCase):
import subprocess import subprocess
import tempfile import tempfile
# To avoid ImportErrors on Windows, where _ctypes does not have
# dlopen and dlsym,
# import here, i.e., inside the test function.
# The skipUnless('linux') decorator ensures that we're on linux
# if we're executing these statements.
from ctypes import CDLL, c_int
from _ctypes import dlopen, dlsym
retcode = subprocess.call(["gcc", "--version"], retcode = subprocess.call(["gcc", "--version"],
stdout=subprocess.DEVNULL, stdout=subprocess.DEVNULL,
stderr=subprocess.DEVNULL) stderr=subprocess.DEVNULL)
@ -111,6 +108,8 @@ class TestNullDlsym(unittest.TestCase):
self.assertEqual(os.read(pipe_r, 2), b'OK') self.assertEqual(os.read(pipe_r, 2), b'OK')
# Case #3: Test 'py_dl_sym' from Modules/_ctypes/callproc.c # Case #3: Test 'py_dl_sym' from Modules/_ctypes/callproc.c
dlopen = test.support.get_attribute(_ctypes, 'dlopen')
dlsym = test.support.get_attribute(_ctypes, 'dlsym')
L = dlopen(dstname) L = dlopen(dstname)
with self.assertRaisesRegex(OSError, "symbol 'foo' not found"): with self.assertRaisesRegex(OSError, "symbol 'foo' not found"):
dlsym(L, "foo") dlsym(L, "foo")
@ -119,5 +118,66 @@ class TestNullDlsym(unittest.TestCase):
self.assertEqual(os.read(pipe_r, 2), b'OK') self.assertEqual(os.read(pipe_r, 2), b'OK')
@unittest.skipUnless(os.name != 'nt', 'test requires dlerror() calls')
class TestLocalization(unittest.TestCase):
@staticmethod
def configure_locales(func):
return test.support.run_with_locale(
'LC_ALL',
'fr_FR.iso88591', 'ja_JP.sjis', 'zh_CN.gbk',
'fr_FR.utf8', 'en_US.utf8',
'',
)(func)
@classmethod
def setUpClass(cls):
cls.libc_filename = find_library("c")
@configure_locales
def test_localized_error_from_dll(self):
dll = CDLL(self.libc_filename)
with self.assertRaises(AttributeError) as cm:
dll.this_name_does_not_exist
if sys.platform.startswith('linux'):
# On macOS, the filename is not reported by dlerror().
self.assertIn(self.libc_filename, str(cm.exception))
@configure_locales
def test_localized_error_in_dll(self):
dll = CDLL(self.libc_filename)
with self.assertRaises(ValueError) as cm:
c_int.in_dll(dll, 'this_name_does_not_exist')
if sys.platform.startswith('linux'):
# On macOS, the filename is not reported by dlerror().
self.assertIn(self.libc_filename, str(cm.exception))
@unittest.skipUnless(hasattr(_ctypes, 'dlopen'),
'test requires _ctypes.dlopen()')
@configure_locales
def test_localized_error_dlopen(self):
missing_filename = b'missing\xff.so'
# Depending whether the locale, we may encode '\xff' differently
# but we are only interested in avoiding a UnicodeDecodeError
# when reporting the dlerror() error message which contains
# the localized filename.
filename_pattern = r'missing.*?\.so'
with self.assertRaisesRegex(OSError, filename_pattern):
_ctypes.dlopen(missing_filename, 2)
@unittest.skipUnless(hasattr(_ctypes, 'dlopen'),
'test requires _ctypes.dlopen()')
@unittest.skipUnless(hasattr(_ctypes, 'dlsym'),
'test requires _ctypes.dlsym()')
@configure_locales
def test_localized_error_dlsym(self):
dll = _ctypes.dlopen(self.libc_filename)
with self.assertRaises(OSError) as cm:
_ctypes.dlsym(dll, 'this_name_does_not_exist')
if sys.platform.startswith('linux'):
# On macOS, the filename is not reported by dlerror().
self.assertIn(self.libc_filename, str(cm.exception))
if __name__ == "__main__": if __name__ == "__main__":
unittest.main() unittest.main()

View file

@ -1,10 +1,11 @@
from test import support
from test.support import import_helper, cpython_only
gdbm = import_helper.import_module("dbm.gnu") #skip if not supported
import unittest
import os import os
from test.support.os_helper import TESTFN, TESTFN_NONASCII, unlink, FakePath import unittest
from test import support
from test.support import cpython_only, import_helper
from test.support.os_helper import (TESTFN, TESTFN_NONASCII, FakePath,
create_empty_file, temp_dir, unlink)
gdbm = import_helper.import_module("dbm.gnu") # skip if not supported
filename = TESTFN filename = TESTFN
@ -205,6 +206,16 @@ class TestGdbm(unittest.TestCase):
self.assertNotIn(k, db) self.assertNotIn(k, db)
self.assertEqual(len(db), 0) self.assertEqual(len(db), 0)
@support.run_with_locale(
'LC_ALL',
'fr_FR.iso88591', 'ja_JP.sjis', 'zh_CN.gbk',
'fr_FR.utf8', 'en_US.utf8',
'',
)
def test_localized_error(self):
with temp_dir() as d:
create_empty_file(os.path.join(d, 'test'))
self.assertRaises(gdbm.error, gdbm.open, filename, 'r')
if __name__ == '__main__': if __name__ == '__main__':

View file

@ -986,16 +986,9 @@ CDataType_in_dll_impl(PyObject *type, PyTypeObject *cls, PyObject *dll,
#ifdef USE_DLERROR #ifdef USE_DLERROR
const char *dlerr = dlerror(); const char *dlerr = dlerror();
if (dlerr) { if (dlerr) {
PyObject *message = PyUnicode_DecodeLocale(dlerr, "surrogateescape"); _PyErr_SetLocaleString(PyExc_ValueError, dlerr);
if (message) {
PyErr_SetObject(PyExc_ValueError, message);
Py_DECREF(message);
return NULL; return NULL;
} }
// Ignore errors from PyUnicode_DecodeLocale,
// fall back to the generic error below.
PyErr_Clear();
}
#endif #endif
#undef USE_DLERROR #undef USE_DLERROR
PyErr_Format(PyExc_ValueError, "symbol '%s' not found", name); PyErr_Format(PyExc_ValueError, "symbol '%s' not found", name);
@ -3812,17 +3805,10 @@ PyCFuncPtr_FromDll(PyTypeObject *type, PyObject *args, PyObject *kwds)
#ifdef USE_DLERROR #ifdef USE_DLERROR
const char *dlerr = dlerror(); const char *dlerr = dlerror();
if (dlerr) { if (dlerr) {
PyObject *message = PyUnicode_DecodeLocale(dlerr, "surrogateescape"); _PyErr_SetLocaleString(PyExc_AttributeError, dlerr);
if (message) {
PyErr_SetObject(PyExc_AttributeError, message);
Py_DECREF(ftuple); Py_DECREF(ftuple);
Py_DECREF(message);
return NULL; return NULL;
} }
// Ignore errors from PyUnicode_DecodeLocale,
// fall back to the generic error below.
PyErr_Clear();
}
#endif #endif
PyErr_Format(PyExc_AttributeError, "function '%s' not found", name); PyErr_Format(PyExc_AttributeError, "function '%s' not found", name);
Py_DECREF(ftuple); Py_DECREF(ftuple);

View file

@ -1579,10 +1579,11 @@ static PyObject *py_dl_open(PyObject *self, PyObject *args)
Py_XDECREF(name2); Py_XDECREF(name2);
if (!handle) { if (!handle) {
const char *errmsg = dlerror(); const char *errmsg = dlerror();
if (!errmsg) if (errmsg) {
errmsg = "dlopen() error"; _PyErr_SetLocaleString(PyExc_OSError, errmsg);
PyErr_SetString(PyExc_OSError, return NULL;
errmsg); }
PyErr_SetString(PyExc_OSError, "dlopen() error");
return NULL; return NULL;
} }
return PyLong_FromVoidPtr(handle); return PyLong_FromVoidPtr(handle);
@ -1595,8 +1596,12 @@ static PyObject *py_dl_close(PyObject *self, PyObject *args)
if (!PyArg_ParseTuple(args, "O&:dlclose", &_parse_voidp, &handle)) if (!PyArg_ParseTuple(args, "O&:dlclose", &_parse_voidp, &handle))
return NULL; return NULL;
if (dlclose(handle)) { if (dlclose(handle)) {
PyErr_SetString(PyExc_OSError, const char *errmsg = dlerror();
dlerror()); if (errmsg) {
_PyErr_SetLocaleString(PyExc_OSError, errmsg);
return NULL;
}
PyErr_SetString(PyExc_OSError, "dlclose() error");
return NULL; return NULL;
} }
Py_RETURN_NONE; Py_RETURN_NONE;
@ -1631,17 +1636,10 @@ static PyObject *py_dl_sym(PyObject *self, PyObject *args)
return PyLong_FromVoidPtr(ptr); return PyLong_FromVoidPtr(ptr);
} }
#ifdef USE_DLERROR #ifdef USE_DLERROR
const char *dlerr = dlerror(); const char *errmsg = dlerror();
if (dlerr) { if (errmsg) {
PyObject *message = PyUnicode_DecodeLocale(dlerr, "surrogateescape"); _PyErr_SetLocaleString(PyExc_OSError, errmsg);
if (message) {
PyErr_SetObject(PyExc_OSError, message);
Py_DECREF(message);
return NULL; return NULL;
}
// Ignore errors from PyUnicode_DecodeLocale,
// fall back to the generic error below.
PyErr_Clear();
} }
#endif #endif
#undef USE_DLERROR #undef USE_DLERROR

View file

@ -8,6 +8,7 @@
#endif #endif
#include "Python.h" #include "Python.h"
#include "pycore_pyerrors.h" // _PyErr_SetLocaleString()
#include "gdbm.h" #include "gdbm.h"
#include <fcntl.h> #include <fcntl.h>
@ -33,6 +34,24 @@ get_gdbm_state(PyObject *module)
return (_gdbm_state *)state; return (_gdbm_state *)state;
} }
/*
* Set the gdbm error obtained by gdbm_strerror(gdbm_errno).
*
* If no error message exists, a generic (UTF-8) error message
* is used instead.
*/
static void
set_gdbm_error(_gdbm_state *state, const char *generic_error)
{
const char *gdbm_errmsg = gdbm_strerror(gdbm_errno);
if (gdbm_errmsg) {
_PyErr_SetLocaleString(state->gdbm_error, gdbm_errmsg);
}
else {
PyErr_SetString(state->gdbm_error, generic_error);
}
}
/*[clinic input] /*[clinic input]
module _gdbm module _gdbm
class _gdbm.gdbm "gdbmobject *" "&Gdbmtype" class _gdbm.gdbm "gdbmobject *" "&Gdbmtype"
@ -91,7 +110,7 @@ newgdbmobject(_gdbm_state *state, const char *file, int flags, int mode)
PyErr_SetFromErrnoWithFilename(state->gdbm_error, file); PyErr_SetFromErrnoWithFilename(state->gdbm_error, file);
} }
else { else {
PyErr_SetString(state->gdbm_error, gdbm_strerror(gdbm_errno)); set_gdbm_error(state, "gdbm_open() error");
} }
Py_DECREF(dp); Py_DECREF(dp);
return NULL; return NULL;
@ -136,7 +155,7 @@ gdbm_length(gdbmobject *dp)
PyErr_SetFromErrno(state->gdbm_error); PyErr_SetFromErrno(state->gdbm_error);
} }
else { else {
PyErr_SetString(state->gdbm_error, gdbm_strerror(gdbm_errno)); set_gdbm_error(state, "gdbm_count() error");
} }
return -1; return -1;
} }
@ -286,7 +305,7 @@ gdbm_ass_sub(gdbmobject *dp, PyObject *v, PyObject *w)
PyErr_SetObject(PyExc_KeyError, v); PyErr_SetObject(PyExc_KeyError, v);
} }
else { else {
PyErr_SetString(state->gdbm_error, gdbm_strerror(gdbm_errno)); set_gdbm_error(state, "gdbm_delete() error");
} }
return -1; return -1;
} }
@ -297,11 +316,12 @@ gdbm_ass_sub(gdbmobject *dp, PyObject *v, PyObject *w)
} }
errno = 0; errno = 0;
if (gdbm_store(dp->di_dbm, krec, drec, GDBM_REPLACE) < 0) { if (gdbm_store(dp->di_dbm, krec, drec, GDBM_REPLACE) < 0) {
if (errno != 0) if (errno != 0) {
PyErr_SetFromErrno(state->gdbm_error); PyErr_SetFromErrno(state->gdbm_error);
else }
PyErr_SetString(state->gdbm_error, else {
gdbm_strerror(gdbm_errno)); set_gdbm_error(state, "gdbm_store() error");
}
return -1; return -1;
} }
} }
@ -534,10 +554,12 @@ _gdbm_gdbm_reorganize_impl(gdbmobject *self, PyTypeObject *cls)
check_gdbmobject_open(self, state->gdbm_error); check_gdbmobject_open(self, state->gdbm_error);
errno = 0; errno = 0;
if (gdbm_reorganize(self->di_dbm) < 0) { if (gdbm_reorganize(self->di_dbm) < 0) {
if (errno != 0) if (errno != 0) {
PyErr_SetFromErrno(state->gdbm_error); PyErr_SetFromErrno(state->gdbm_error);
else }
PyErr_SetString(state->gdbm_error, gdbm_strerror(gdbm_errno)); else {
set_gdbm_error(state, "gdbm_reorganize() error");
}
return NULL; return NULL;
} }
Py_RETURN_NONE; Py_RETURN_NONE;

View file

@ -320,6 +320,7 @@ _setException(PyObject *exc, const char* altmsg, ...)
va_end(vargs); va_end(vargs);
ERR_clear_error(); ERR_clear_error();
/* ERR_ERROR_STRING(3) ensures that the messages below are ASCII */
lib = ERR_lib_error_string(errcode); lib = ERR_lib_error_string(errcode);
func = ERR_func_error_string(errcode); func = ERR_func_error_string(errcode);
reason = ERR_reason_error_string(errcode); reason = ERR_reason_error_string(errcode);

View file

@ -134,6 +134,7 @@ _pysqlite_seterror(pysqlite_state *state, sqlite3 *db)
/* Create and set the exception. */ /* Create and set the exception. */
int extended_errcode = sqlite3_extended_errcode(db); int extended_errcode = sqlite3_extended_errcode(db);
// sqlite3_errmsg() always returns an UTF-8 encoded message
const char *errmsg = sqlite3_errmsg(db); const char *errmsg = sqlite3_errmsg(db);
raise_exception(exc_class, extended_errcode, errmsg); raise_exception(exc_class, extended_errcode, errmsg);
return extended_errcode; return extended_errcode;

View file

@ -3,6 +3,7 @@
#include "parts.h" #include "parts.h"
#include "util.h" #include "util.h"
#include "clinic/exceptions.c.h" #include "clinic/exceptions.c.h"

View file

@ -374,6 +374,7 @@ pymain_run_file_obj(PyObject *program_name, PyObject *filename,
if (fp == NULL) { if (fp == NULL) {
// Ignore the OSError // Ignore the OSError
PyErr_Clear(); PyErr_Clear();
// TODO(picnixz): strerror() is locale dependent but not PySys_FormatStderr().
PySys_FormatStderr("%S: can't open file %R: [Errno %d] %s\n", PySys_FormatStderr("%S: can't open file %R: [Errno %d] %s\n",
program_name, filename, errno, strerror(errno)); program_name, filename, errno, strerror(errno));
return 2; return 2;

View file

@ -1782,7 +1782,12 @@ add_error(PyObject *errors_module, PyObject *codes_dict,
* with the other uses of the XML_ErrorString function * with the other uses of the XML_ErrorString function
* elsewhere within this file. pyexpat's copy of the messages * elsewhere within this file. pyexpat's copy of the messages
* only acts as a fallback in case of outdated runtime libexpat, * only acts as a fallback in case of outdated runtime libexpat,
* where it returns NULL. */ * where it returns NULL.
*
* In addition, XML_ErrorString is assumed to return UTF-8 encoded
* strings (in conv_string_to_unicode, we decode them using 'strict'
* error handling).
*/
const char *error_string = XML_ErrorString(error_code); const char *error_string = XML_ErrorString(error_code);
if (error_string == NULL) { if (error_string == NULL) {
error_string = error_info_of[error_index].description; error_string = error_info_of[error_index].description;

View file

@ -299,6 +299,15 @@ PyErr_SetString(PyObject *exception, const char *string)
_PyErr_SetString(tstate, exception, string); _PyErr_SetString(tstate, exception, string);
} }
void
_PyErr_SetLocaleString(PyObject *exception, const char *string)
{
PyObject *value = PyUnicode_DecodeLocale(string, "surrogateescape");
if (value != NULL) {
PyErr_SetObject(exception, value);
Py_DECREF(value);
}
}
PyObject* _Py_HOT_FUNCTION PyObject* _Py_HOT_FUNCTION
PyErr_Occurred(void) PyErr_Occurred(void)